Search

Background: The Center for Medicare and Medicaid Services uses primary discharge ICD-10 codes for Acute Myocardial Infarction (AMI) to define inclusion in value-based payment programs such as the hospital readmissions reduction program. Prior studies have shown that a significant number of hospitalizations with an ICD-10 code for AMI actually represent a Type 2 MI (T2MI) or non-ischemic myocardial injury (NIMI). Yet, the care of such patients may differ from those with AMI, and their inclusion in value-based payment programs risks inaccurately assessing hospital care and even unwarranted financial penalties. In October 2018, a new ICD-10 code for T2MI was introduced (I21.A1). It is unclear what effect, if any, this new code has had on Type 1 MI (T1MI) representation among discharges assigned an ICD-10 code for AMI. Methods: We screened all discharges from 8 UCHealth hospitals between January 1, 2017 and June 30, 2019 for a primary discharge ICD-10 code of AMI (n=3,445). Hospitalizations occurring during a six month period prior to the introduction of I21.A1 (January 1, 2017 to June 30, 2017) and a six-month period after the introduction of I21.A1 (January 1, 2019 to June 30, 2019) were selected for chart review. Using the 4th Universal Definition of MI, hospitalizations were assigned a gold-standard clinical diagnosis of T1MI, T2MI, or NIMI. The cohorts were compared according to the primary outcomes of readmission and mortality at 30 and 90 days. Results: Of 1,364 discharges reviewed, 676 occurred in the pre-I21.A1 period and 667 occurred in the post-I21.A1 period. In the pre-I21.A1 cohort, 86% (585 of 676) of discharges met clinical criteria for T1MI, 10% (67 of 676) were deemed T2MI and 4% (24 of 676) represented NIMI. In the post-I21.A1 group, 93% (626 of 676) were deemed T1MI, while 5% (33 of 667) were deemed T2MI and 2% (12 of 667) met criteria for NIMI. The post-I21.A1 cohort had significantly fewer readmissions at 30 and 90 days when compared to the pre-I21.A1 cohort, but had similar rates of mortality. Conclusions: After the introduction of an ICD-10 code for T2MI, fewer patients with T2MIs and NIMIs were eligible for inclusion in value-based programs. Concomitantly, we observed lower readmission rates among patients with an ICD-10 code for AMI after the T2MI code was introduced.

In the United States, the positive predictive value (PPV) of cardiac troponin for type 1 myocardial infarction is substantially lower than in Europe (15% vs. 50%). Further, even with publication of the 4 th Universal Definition of Myocardial Infarction, recent studies have shown that inaccurate classification of myocardial injury is common among clinicians in the United States. These findings are at least partly attributable to clinicians’ knowledge and attitudes about cardiac troponin testing; a survey of these parameters has never been conducted. Clinicians at the University of Colorado completed a brief 8-question multiple-choice survey related to troponin use, definitions of myocardial infarction and clinical assessment of elevated troponin levels. The survey was distributed via secure email and administered electronically using the Qualtrics™ platform. Responses were anonymous, completion was estimated to take 3 minutes and a lottery award system was used as an incentive for participation. Respondents included trainees, advanced practice providers and attending physicians from internal medicine, emergency medicine and medical subspecialties. We plan to obtain a total of 300 responses with descriptive findings of preliminary results included below. The survey was completed by 114 clinicians: 37 interns (32%), 45 residents (39%), 9 advanced practice providers (8%), 11 fellows (10%), and 12 attending physicians (11%). Regarding indications for troponin testing, 93% (106/114) indicated that they “usually” or “always” check troponin levels in patients with chest pain. More interestingly, 46% (52/112) reported checking troponin on “undifferentiated patients” at least half the time. For troponin interpretation, 97% (110/114) of participants identified that troponin levels alone cannot rule in or rule out coronary artery disease. In contrast, only 36% (41/114) and 55% (63/114), respectively, identified the NPV and PPV of a contemporary troponin assay for type 1 MI. Further, only 50% (57/114) of respondents identified that the likelihood of type 1 MI increases as troponin levels increase. Three brief clinical vignettes revealed that, while 78% (89/114) and 74% (45/61) of participants, respectively, identified type 1 MI and type 2 MI presentations, only 40% (21/53) of respondents correctly identified a vignette for non-ischemic myocardial injury. Concordant with this finding, 54% (61/114) of clinicians correctly identified the 4 th Universal Definition of Myocardial Infarction. These preliminary findings highlight important facets of clinician attitudes and knowledge about troponin testing that help explain the poor PPV for troponin and diagnostic misclassification observed among U.S. clinicians. These results could help guide curricular and clinical decision support interventions designed to improve the use and interpretation of cardiac troponin testing.

Background: Patients with elevated troponin (cTn) in the absence of Acute Myocardial Infarction (AMI) present a diagnostic dilemma: they at high risk of adverse cardiac events, yet there is little evidence on how they should be managed clinically. In recent years, greater sensitivity of cTn assays and less selective use of cTn in clinical practice have led to a greater number of non-AMI patients with elevated troponin. The exact scope of the problem in routine clinical practice is not known, as published estimates of troponin’s positive predictive value (PPV) for AMI range from 15% to 70%. We sought to define troponin’s PPV for AMI at our institution and examine both process and outcome measures for AMI and non-AMI patients. Methods: In a retrospective cohort analysis of patients evaluated at the University of Colorado Hospital, we identified 5,903 hospital encounters between January 1, 2017 and October 6, 2018 in which patients were found to have an elevated serum troponin level (>0.04 ng/mL, Siemens Advia Centaur Tnl-Ultra). We used ICD-10 codes, billing diagnosis related groups (DRGs), or inclusion in the NCDR® ACTION Registry® to identify patients with a diagnosis of AMI. Patients not captured by one of these diagnoses were categorized as non-AMI. We then compared AMI and non-AMI patients according to the primary outcome of 100-day mortality. Secondary outcomes and process measures were also examined. Results: Out of 5,903 hospital encounters in which elevated cTn was detected, 730 were associated with any diagnosis of AMI (PPV 12.4%). The PPV was lower for individual AMI diagnosis groups. The primary outcome of 100-day mortality was observed in 10.3% of AMI patients versus 20.5% of non-AMI patients (p < 0.001). AMI patients also had significantly shorter hospitalizations, higher rates of inpatient echocardiography and higher rates of P2Y12 inhibitor therapy (Table 1). Conclusion: The PPV of a contemporary troponin assay for AMI was 12.4% in routine clinical practice at a tertiary care academic hospital. Concordant with prior studies, non-AMI patients had worse outcomes. With high-sensitivity troponin (hsTn) assays poised to increase the prevalence of non-AMI troponin elevation, guidance is needed regarding the appropriate use of hsTn as well as the evaluation and treatment of non-ACS patients.

Background: In the Unites States, there are three major approaches to identify patients diagnosed with Acute Myocardial Infarction (AMI). The Center for Medicare and Medicaid Services uses ICD-10 codes (1) for the Value-Based Payments program whereas it uses DRG codes (2) for the Inpatient Prospective Payment System. Hospital administrators, meanwhile, often use the NCDR® ACTION Registry® (3) to examine process and outcome measures for their AMI patients. Understanding which patients are captured by these different classification systems, and whether there is agreement between them, has important implications for hospitals’ efforts to optimize performance measures and payments for AMI. Methods: In a retrospective cohort analysis, we examined 730 hospitalizations at the University of Colorado Hospital between January 1, 2017 and October 6, 2018 that were captured by one of the three methods for identifying AMI patients: 1) ICD-10, 2) DRG, or 3) ACTION. Categories were not mutually exclusive. Agreement between DRG, ICD-10 and ACTION diagnoses of AMI was assessed by percentage agreement with the total AMI cohort and dual-comparison kappa statistics. AMI cohorts defined by ICD-10 vs. DRG vs. ACTION were then compared according to the primary outcome of 100-day mortality, as well as secondary outcomes of 30-day readmission, length of stay (LOS), P2Y12 inhibition and rates of inpatient echocardiography. Results: Among 730 hospitalizations with at least one diagnosis of AMI, 617 (84.5%) were assigned a primary ICD-10 code for AMI, 227 (31.1%) were assigned a DRG code for AMI, and 479 (65.6%) carried a diagnosis of AMI in ACTION. Only 112 (15.3%) were captured by all 3 diagnosis groups (ICD-10, DRG and ACTION). 159 (21.7%) hospitalizations were captured by both ICD-10 and DRG diagnoses (Kappa -0.143); 425 (58.2%) were included in ICD-10 and ACTION groups (Kappa 0.141); 121 (16.6%) were included in DRG and ACTION groups (Kappa -0.137). Outcomes and process measures as a function of diagnosis category are presented in Figure 1. Conclusion: Only about 15% of patients with a diagnosis of AMI were included in all three diagnosis groups (ICD-10, DRG and ACTION). There may be significant between-group differences in outcomes and process measures. These data emphasize the need to harmonize administrative definitions for AMI.